set consistent field office name

Signed-off-by: John Seekins <[email protected]>

Author: johnseekins

enricher.py

@@ -32,8 +32,6 @@ def enrich_facility_data(self, facilities_data: dict) -> dict:
             facility_name = facility["name"]
             logger.info("Processing facility %s/%s: %s...", processed + 1, total, facility_name)
             enriched_facility = copy.deepcopy(facility)
-            if not enriched_facility["field_office"]:
-                enriched_facility["field_office"] = "(Possibly) Not managed by DHS field office"
 
             # Wikipedia search # todo refactor to method
             try:

scraper.py

@@ -331,9 +331,17 @@ def scrape_facilities(self):
                         if url in self.facilities_data["facilities"][full_address]["source_urls"]:
                             continue
                         self.facilities_data["facilities"][full_address]["source_urls"].append(url)
+                    if not self.facilities_data["facilities"][full_address].get("field_office", ""):
+                        self.facilities_data["facilities"][full_address]["field_office"] = (
+                            "(Possibly) Not managed by DHS field office"
+                        )
                 # this is likely to produce _some_ duplicates, but it's a reasonable starting place
                 else:
                     self.facilities_data["facilities"][facility["name"]] = facility
+                    if not self.facilities_data["facilities"][facility["name"]].get("field_office", ""):
+                        self.facilities_data["facilities"][facility["name"]]["field_office"] = (
+                            "(Possibly) Not managed by DHS field office"
+                        )
 
         self.facilities_data["scrape_runtime"] = time.time() - start_time
         logger.info("Total facilities scraped: %s", len(self.facilities_data["facilities"]))